\documentclass[12pt]{article}
%\usepackage{lmodern}
%\usepackage[T1]{fontenc}
%\usepackage[latin9]{inputenc}
\usepackage[sc]{mathpazo}
%\usepackage[english]{babel}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{hyperref}
\usepackage{setspace}
\usepackage{adjustbox}
\usepackage{amsmath}
\usepackage{graphicx}
\usepackage[colorinlistoftodos]{todonotes}
\usepackage{tabu}
%\usepackage[a4paper, total={6.4in, 9.4in}]{geometry}
%\usepackage[margin=1in]{geometry}
\usepackage[letterpaper]{geometry}
\geometry{verbose,tmargin=2.0cm,bmargin=2.0cm,lmargin=2.5cm,rmargin=2.5cm}
\usepackage{mathptmx}
\usepackage{adjustbox}
\usepackage{subcaption}
\usepackage{xcolor}
\hypersetup{
	colorlinks,
	linkcolor={blue!90!black},
	citecolor={blue!90!black},
	urlcolor={blue!90!black}
}



\begin{document}
	\onehalfspacing
	\begin{titlepage}
	\title{\Large{\bf Employment Protection Legislation and Informality \\ Theory and Evidence from India}}  \vspace{.75em}
	\date{REPLICATION OF RESULTS}
	\maketitle
	\begin{center}
		{\large \hspace{2mm}  Ritam Chaurey \hspace{8mm} Gaurav Chiplunkar  \hspace{8mm} Vidhya Soundararajan}
	\end{center}
\end{titlepage}

\newpage

%******************************
% MAIN FIGURES
%******************************

\section{Figures in the Main Paper}

%Figure 1
\begin{figure}[ht!]
\begin{center}
\caption{Theoretical Implications of Penalizing Hiring Contract Workers \label{fig:Bc_inf}} \label{subsubsec: BcTax}
\begin{subfigure}{0.45\textwidth}
    \centering
	\includegraphics[scale=0.45]{calibration/Figure1a.png}
    \caption{Margins of Informality}\label{fig:Bc_inf_a}
\end{subfigure}
\hfill
\begin{subfigure}{0.5\textwidth}
    \centering
	\includegraphics[scale=0.45]{calibration/Figure1b.png}
    \caption{Productivity Thresholds}\label{fig:Bc_zstar}
\end{subfigure}
\hfill 
 % \bigskip 
\end{center}
\raggedright
\footnotesize
\textit{\underline{Note:}} The above figure examines the impact of penalizing the hiring of contract workers on the intensive and extensive margins of informality (Graph A) and on the productivity of the marginal firm, which is indifferent between entering the informal and formal sectors (Graph B). We start with a baseline scenario of $b_c=1$, and examine the impact of gradually increasing $b_c$. We use the following parameter values for the simulation: $b_p =1$; $E_{I}=0.5$; $E_{R}=50$; $\tau=0.05$; $\sigma_x= \sigma_{\varepsilon} = 0.2$; $\nu = 3$; $\phi = 1$;  $\rho=0.74$; $\theta=1.15$; $N=L=5m$.
\end{figure}


%Figure 2
\begin{figure}[h!]
	\begin{center}
		\caption{Impact of the Policy on the Use of Contract and Regular Workers}
		\label{fig:eventstudy}
		\begin{subfigure}{.4\linewidth}
			\centering
			\includegraphics[width=8cm]{asi/Figure2a.png}
			\caption{Log Contract Workers}\label{fig:lncw}
		\end{subfigure}
		\hfill
		\hfill 
		\begin{subfigure}{.4\linewidth}
			\centering
			\includegraphics[width=8cm]{asi/Figure2b.png}
			\caption{Log Regular Workers}\label{fig:lnrw}
		\end{subfigure}
		\\ 
		\begin{subfigure}{.4\linewidth}
			\centering
			\includegraphics[width=8cm]{asi/Figure2c.png}
			\caption{Frac. Contract Workers}\label{fig:cwratio}
		\end{subfigure}
	\end{center}
	% \bigskip 
	\footnotesize
	\raggedright
	\textit{\underline{Notes:}} The above figure uses establishment-level data from the Annual Survey of Industries between 1999-2005. The graphs plot the regression coefficients from a difference-in-differences specification presented in  \autoref{tab:eventstudy_regressionoutput}. The black squares report the results for workers, while the red diamonds report the results for worker-days. All regressions contain firm  and 2-digit industry-year fixed effects, along with a state-time trend. 90\% confidence intervals, clustered at the state level are indicated around the point estimate. The coefficient for 2002-2003, the year before the reform, has been normalized to zero. The outcome variables are winsorized at the top and the bottom 1\% of the distribution.
\end{figure}





%******************************
% MAIN TABLE
%******************************
\newpage
\section{Tables in the Main Paper}

% Table 1
\begin{table}[h!]
	\caption{Summary Statistics from the ASI Data}
	\label{tab:summarystats}
	\input{asi/Table1.tex}
	\raggedright  \footnotesize \vspace{0.2cm} \\ 
	\textit{\underline{Notes:}} The above table uses data from the Annual Survey of Industries between 1999-2002. Each row reports an outcome variable. Columns (1)-(3) report the total number of observations, mean, and standard deviation for the variable in Andhra Pradesh, while Columns (4)-(6) report the same for all other Indian states. All variables have been windsorized at the top and bottom 1\%.
\end{table}

	
	
% Table 2
\newpage
\begin{table}[t!]
	\caption{Overlap Across Measures of Informality}
	\label{tab:summarystats_nss}
	\begin{center}
		\centering
	\input{nss/Table2}
	\end{center}
	\raggedright \footnotesize 
	\textit{\underline{Notes:}} The above table uses data rounds of the National Sample Survey between 1999-2005. Each outcome variable takes the value 1 if an individual reports working on an informal contract (Column 1), in an informal firm (Column 2), or in casual work (Column 3). Each panel reports the fraction of individuals in each of the above categories if a worker is on an informal contract (Row 1), informal firm (Row 2), and casual work (Row 3). See Appendix Table A1 for detailed definitions for each variable.
\end{table}







% Table 3
\clearpage \newpage
\begin{table}[h!]
	\caption{Impact of the Policy on Contract, Payroll and Total Workers in Formal Sector Firms} 
	\label{tab:mainresults} \label{tab:regulartotalworkers}
	\centering
	\adjustbox{max width = \linewidth}{
		\input{asi/Table3.tex}}
	\\ \vspace{0.1cm} 
	\footnotesize \raggedright \textit{\underline{Notes:}} The above table uses establishment-level data from the Annual Survey of Industries between 1999-2005. It reports the impact of the policy reform on Workers in Columns (1) and (2) and Worker-Days in Columns (3) and (4). Panel A reports the impact of the policy on the usage of contract workers, while Panel B reports the impact on the usage of regular payroll workers and total workers. Frac. Contract is the number of contract workers (or worker-days) as a fraction of total workers (worker-days) in a firm. Post is defined as 1 for years after 2003, and 0 before that. Treat is defined as 1 for Andhra Pradesh and 0 for other states. The outcome variables are winsorized from the top and bottom at 1\%. All regressions contain firm and 2-digit industry-year fixed effects, a state-time trend, and control for age and age-square of the firm. Robust standard errors clustered at the state level in parentheses. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. 
\end{table}




% Table 4
\clearpage \newpage
\begin{table}[h!]
	\caption{Impact of the Policy on the Probability of Working in the Informal Sector and Daily Wages}\label{table:informalNSS}
	\input{nss/Table4}
	\raggedright \footnotesize \\ \vspace{0.1cm} 
	\textit{\underline{Notes:}} The above table uses data from all rounds of the National Sample Survey (NSS) between 1999-2005. Each outcome variable takes the value 1 if an individual reports working on an informal contract (Column 1), informal firm (Column 2), and casual work (Column 3). Daily log wages are reported in Column (4) and windsorized at the top and bottom 1\% of the distribution. \autoref{tab:datadef} provides detailed definitions for all outcome variables. Post is a binary variable that takes the value 1 for years after 2003, while Treat is a binary variable that takes the value 1 for Andhra Pradesh and 0 otherwise. All regressions contain state-2 digit industry and 2-digit industry-year fixed effects along with a state-time trend. They also control for individual characteristics such as age, age-squared, gender, literacy, caste group, and marital status. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. We use sample weights for the estimation and report robust standard errors clustered at the state level in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. The sample size varies depending on the coverage of the underlying variable across rounds as well as whether a worker reports it.
\end{table}



%Table 5
\begin{table}[t!]
\caption{Impact of the Probability of Firms Remaining Unregistered}\label{table:informalEC}
\begin{center}
\centering
\input{ec/Table5}
\end{center}
\raggedright \footnotesize \vspace{0.1cm}
\textit{\underline{Notes:}} The above table uses data from the 1998 and 2005 rounds of the Economic Census. The outcome variable is a binary variable that takes the value 1 if a firm is unregistered and 0 otherwise. We classify firms as Large (Column 1) if they employ more than 10 workers, and Small (Column 2) if they employ less than 10 workers. Column 3 includes all firms. All regressions contain district, year, and state-2-digit industry fixed effects and control for firm characteristics like whether the firm is owned by a female, whether the owner is from a lower-caste, and whether the firm uses electrical power. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. 
\end{table}



%Table 6
\clearpage \newpage
\begin{table}[h!]
	\caption{Entry-Exit into the Formal Sector \label{tab:asi_entryexit}} 
	\begin{center}
		\centering
		\input{asi/Table6}
	\end{center}
\raggedright \footnotesize \vspace{0.1cm}  
\textit{\underline{Notes:}} Data are from the repeated cross-section of the Annual Survey of Industries between 1999-2005 that have been aggregated to the state-3-digit industry level. Column (1) reports the fraction of firms from year $t-1$ that become inactive in year $t$, while Column (2) reports the number of firms that enter the ASI sample in year $t$ as a fraction of firms in $t-1$. Post is defined as 1 for years after 2003, and 0 before that. Treat is defined as 1 for Andhra Pradesh, and 0 for other states. Robust standard errors are clustered at the state level and reported in parentheses. * is $p < 0.1$, ** is $p < 0.05$, *** is $p < 0.01$.
\end{table}




%Table 7
\clearpage \newpage
\begin{table}[h!]
	\caption{Impact of Employment Protection Policies}\label{table:counterfactual_bc}
	\input{calibration/Table7}
\raggedright \footnotesize
\textit{\underline{Notes:}} Notes: The above table reports the impact of two labor laws on the margins of informality. The baseline values in Column (1) have been normalized to 1, except for the productivity of the marginal firm in the formal sector, which in the baseline has been normalized to 1 for the productivity of the marginal firm in the informal sector. Columns (2) and (3) report the impact of a 19\%  (12\%) increase (decrease) in $b_c$ ($b_r$) relative to its baseline value. The change in $b_r$ has been calibrated to achieve the same reduction in the fraction of contract workers as in Columns (2).
\end{table}




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% APPENDIX FIGURES
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newpage 
\setcounter{table}{0}
\setcounter{figure}{0}
\renewcommand{\thetable}{A\arabic{table}}
\renewcommand{\thefigure}{A\arabic{figure}}
\setcounter{section}{0}
\renewcommand{\thesection}{A\arabic{section}}
\renewcommand{\thesection}{\Alph{section}}
\newgeometry{left=0.8in}


\section{Appendix Figures and Tables}


% Figure A1
\begin{figure}[htbp]
\begin{center}
\caption{Fraction of Informal Workers and EPL Enforcement \label{fig:Inf}}
\begin{subfigure}[b]{0.4\textwidth}
    \centering
	\includegraphics[scale=0.35]{global/FigureA1_a.png}
    \caption{Fraction of Informal Workers}
    \label{fig:Inf_workers}
\end{subfigure}
\hfill 
\begin{subfigure}[b]{0.4\textwidth}
    \centering
	\includegraphics[scale=0.35]{global/FigureA1_b.png}
    \caption{EPL Enforcement}
    \label{fig:Inf_EPL}
\end{subfigure}
 % \bigskip 
\end{center}
\footnotesize
\raggedright
\textit{\underline{Note:}} Data on informality is taken from Elgin, Kose, Ohnsorge, and Yu (2021). Data on enforcement of labor regulation is taken from Botero et al. (2004). We use 2018 values across all countries. Data on Real GDP per-capita is taken from the World Bank.
\end{figure}



% Figure A2
\begin{figure}[ht!]
	\begin{center}
		\caption{Fraction of Workers and Firms in the Informal Sector}
		\label{fig:ec_inf_ap}
		\includegraphics[width = \textwidth, scale=0.25]{ec/FigureA2.png}
	\end{center}
	\raggedright 
	\footnotesize
	\textit{\underline{Note:}} The above figure uses data from the 1998 round of the Economic Census. The blue bars plot the fraction of unregistered firms in each industry, while the red bars are the fraction of workers who work in them. A.P. = Andhra Pradesh; Control = Other states in India excluding Andhra Pradesh.
\end{figure}



%Figure A3
\begin{figure}[ht!]
	\begin{center}
		\caption{Impact of the Policy on the Use of Contract and Regular Workers
			\label{fig:append_eventstudy}}
		\begin{subfigure}{.4\linewidth}
			\centering
			\includegraphics[width=8cm]{asi/FigureA3a.png}
			\caption{Log Contract Workers}\label{fig:append_lncw}
		\end{subfigure}
		\hfill
		\hfill 
		\begin{subfigure}{.4\linewidth}
			\centering
			\includegraphics[width=8cm]{asi/FigureA3b.png}
			\caption{Log Regular Workers}\label{fig:append_lnrw}
		\end{subfigure}
		\\ 
		\begin{subfigure}{.4\linewidth}
			\centering
			\includegraphics[width=8cm]{asi/FigureA3c.png}
			\caption{Frac. Contract Workers}\label{fig:append_cwratio}
		\end{subfigure}
	\end{center}
	% \bigskip 
	\footnotesize
	\raggedright
	\textit{\underline{Notes:}} The above graphs plot the regression coefficients from a difference-in-differences specification using data from the Annual Survey of Industries between 1999-2008. The black squares report the results for workers, while the red diamonds report the results for worker-days. 90\% confidence intervals, clustered at the state level are indicated around the point estimate. The coefficient for 2002-2003, the year before the reform, has been normalized to zero. The outcome variables are winsorized at the 1\% levels at the top and the bottom of the distribution.
\end{figure}




%Figure A4
\begin{figure}[ht!]
	\centering
	\caption{Correlation between the Fraction of Contract Workers and Firm-Size in Andhra Pradesh}
	\includegraphics[width=\linewidth]{calibration/FigureA4.png} \label{fig:firmsize_corr}
	\raggedright \footnotesize \vspace{0.1cm} \\ 
	\textit{\underline{Notes:}} The above figure uses establishment-level data from the Annual Survey of Industries for Andhra Pradesh in 2002 (before the policy) and plots a non-parameteric correlation (solid black line) between log firm-size on the horizontal axis and the fraction of contract workers on the vertical axis. The dotted gray lines are the 95\% confidence intervals. The dash-dotted blue line is the fit simulated from the model. 
\end{figure}



\setcounter{figure}{0}
\renewcommand{\thefigure}{C\arabic{figure}}

%Figure C1
\begin{figure}[ht!]
	\begin{center}
		\caption{Distribution of Contract Workers Across Occupations in Formal Firms} 
		\label{fig:nss_contract}
		\includegraphics[scale=0.45]{nss/FigureC1.png}
	\end{center}
	\raggedright
	\footnotesize 
	\textit{\underline{Note:}} The above figure pools the sample of workers working in the formal sector across multiple rounds of the NSS between 1999-00 and 2009-10. The figure plots the fraction of contract workers across occupations. The dotted gray lines report the 25th and 75th percentiles.
\end{figure}

%Figure C2
\begin{figure}[h]
		\caption{Size-based Penalty Function \label{fig:Probdetection}}
        \centering 
		\includegraphics[scale=0.7]{calibration/FigureC2.png}
		
	\small \raggedright \textit{\underline{Notes:}} The above graph plots the size-based penalty function of operating in the informal sector, as a function of firm size.
\end{figure}



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% APPENDIX TABLES
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newpage 

% Table A1
\begin{table}[ht!]
	\caption{Definition of Variables, Coverage of Data Over Time}
	\label{tab:datadef}
	\input{asi/TableA1}
\end{table}



% Table A2
\begin{table}[h!]
	\caption{Event Study Regression \label{tab:eventstudy_regressionoutput}} 
	\centering
	\adjustbox{max width = \linewidth}{\input{asi/TableA2.tex}}
	\\ \footnotesize \raggedright \vspace{0.1cm}
	\textit{\underline{Notes:}} Data are from the Annual Survey of Industries between 1999-2005. Columns (1)-(6) report underlying coefficients in \autoref{fig:eventstudy} for sub-figures (a)-(c) respectively. The coefficient for 2002-2003, the year before the reform, has been normalized to zero. The outcome variables are winsorized at the 1\% levels at the top and the bottom of the distribution. Following Borusyak et al. (2024), the p-value of a F-test that jointly tests for the pre-policy $\widehat{\beta_t}$ to be equal to 0 are reported. Robust standard errors clustered at the state level in parentheses; *** is $p<0.01$, ** is $p<0.05$ and * is $p<0.1$.  
\end{table}




% Table A3
\begin{table}[h!]
	\caption{Impact on Core and Non-Core Activities \label{tab:asi_corenoncore}} 
	\adjustbox{max width = \linewidth}{
	\input{asi/TableA3}}
	\raggedright \footnotesize \vspace{0.1cm} \\ 
	\textit{\underline{Notes:}} Data are from the Annual Survey of Industries between 1999-2005. Core activity (Panel A) is any activity for which the establishment is set up, and other activities essential for these core activities. Non-core activities (Panel B) are the remaining peripheral activities (listed in the Appendix Section B). Column (1)-(3) reports the results for contract workers, regular workers, and the fraction of contract workers respectively. Frac. Contract is the number of contract workers (or worker-days) as a fraction of total workers (worker-days) in a firm. Post is defined as 1 for years after 2003, and 0 before that. Treat is defined as 1 for Andhra Pradesh, and 0 for other states. The outcome variables are winsorized from the top and bottom at 1\%. Robust standard errors are clustered at the state level and reported in parentheses. * is $p < 0.1$, ** is $p < 0.05$, *** is $p < 0.01$.
\end{table}







% Table A4
\begin{table}[ht!]
	\caption{Results After Restricting the Control States to Neighboring States Only \label{tab:asi_neighbor}} 
	\input{asi/TableA4}
	\raggedright \footnotesize \vspace{0.1cm} \\ 
	\textit{\underline{Notes:}} The above table uses data from the Annual Survey of Industries between 1999-2005. The sample is constrained to neighboring states of Andhra Pradesh, which are Chhattisgarh, Karnataka, Maharashtra, and Odisha. It reports the impact of the policy reform on workers in Columns (1) and (2) and worker-Days in Columns (3) and (4). Panel A reports the impact of the policy on the usage of contract workers, while Panel B reports the impact on the usage of regular payroll workers and total workers. Frac. Contract is the number of contract workers (or worker-days) as a fraction of total workers (worker-days) in a firm. Post is defined as 1 for years after 2003, and 0 before that. Treat is defined as 1 for Andhra Pradesh, and 0 for other states. The outcome variables are winsorized from the top and bottom at 1\%. Robust standard errors are clustered at the state level and reported in parentheses. * is $p < 0.1$, ** is $p < 0.05$, *** is $p < 0.01$.
	
\end{table}



% Table A5
\begin{table}[h!]
	\caption{Results After Excluding Neighboring States  \label{tab:asi_noneighbor}} 
	\adjustbox{max width = \linewidth}{
	\input{asi/TableA5}}
	\raggedright \footnotesize \vspace{0.1cm} \\ 
\textit{\underline{Notes:}} The above table uses data from the Annual Survey of Industries between 1999-2005. The sample excludes neighboring states of Andhra Pradesh, which are Chhattisgarh, Karnataka, Maharashtra, and Odisha. It reports the impact of the policy reform on workers in Columns (1) and (2) and worker-Days in Columns (3) and (4). Panel A reports the impact of the policy on the usage of contract workers, while Panel B reports the impact on the usage of regular payroll workers and total workers. Frac. Contract is the number of contract workers (or worker-days) as a fraction of total workers (worker-days) in a firm. Post is defined as 1 for years after 2003, and 0 before that. Treat is defined as 1 for Andhra Pradesh, and 0 for other states. The outcome variables are winsorized from the top and bottom at 1\%. Robust standard errors are clustered at the state level and reported in parentheses. * is $p < 0.1$, ** is $p < 0.05$, *** is $p < 0.01$.
\end{table}





% Table A6
\begin{table}[t!]
	\begin{center}
		\caption{Impact of the Policy Using Synthetic Control Weights} 
		\centering  
		\input{asi/TableA6}
		\label{tab:append_synthweights}
	\end{center}
	\raggedright \footnotesize 
	\textit{\underline{Notes:}} The above table uses establishment-level data from the Annual Survey of Industries from 2000-2006. Regressions are weighted by synthetic control weights generated by matching the treated unit with a weighted combination of control units by matching on regular workers, non-literate population, scheduled caste population, and cultivators population in the years 2000 and 2001. The above table reports the impact of the policy reform on Workers in Columns (1) and (2) and Worker-Days in Columns (3) and (4). Panel A reports the impact of the policy on the usage of contract workers, while Panel B reports the impact on the usage of regular payroll workers and total workers. Frac. Contract is the number of contract workers (or worker-days) as a fraction of total workers (worker-days) in a firm. Post is defined as 1 for years after 2003, and 0 before that. Treat is defined as 1 for Andhra Pradesh, and 0 for other states. Source of the data is the Annual Survey of Industries between 2000-2007. The outcome variables are winsorized from the bottom at 1\% and from the top at 99\%. Robust standard errors clustered at the state level in parentheses. All regressions contain firm and 2-digit industry-year fixed effects, a state-time trend, and control for age and age-square of the firm. Robust standard errors clustered at the state level in parentheses. * is p$<$0.1, ** is p$<$0.05, *** is p$<$0.01. 
\end{table}






% Table A7
\begin{table}[h!]
	\caption{Robustness of ASI Results to Alternate Fixed Effects \label{tab:asi_fe}} 
	\adjustbox{max width = \linewidth}{
		\input{asi/TableA7}}
	\raggedright \footnotesize \vspace{0.1cm} \\ 
	\textit{\underline{Notes:}} The above table uses data from the Annual Survey of Industries between 1999-2005. The outcome variables in Columns (1)-(2) are log contract worker and worker-days respectively; Columns (3)-(4) are log regular workers and worker-days respectively, and Columns (5)-(6) are the fraction of contract workers and worker-days respectively. Regressions in Panel A include year fixed effects and a state-time trend. Panels B and C additionally include 2-digit industry-year fixed effects, and state-2-digit industry fixed effects respectively. Panel D includes both state-2-digit-industry and 2-digit industry-year fixed effects. Lastly, Panel E includes firm fixed effects and 2-digit industry-year fixed effects, as in the specification in the paper. Post is defined as 1 for years after 2003, and 0 before that. Treat is defined as 1 for Andhra Pradesh, and 0 for other states. The outcome variables are winsorized from the top and bottom at 1\%. Robust standard errors are clustered at the state level and reported in parentheses. * is $p < 0.1$, ** is $p < 0.05$, *** is $p < 0.01$.
\end{table}




% Table A8
\begin{table}[ht!]
	\caption{Robustness of Results to Not Using Sampling Weights \label{tab:nss_wowts}} 
\begin{center}
		\input{nss/TableA8}
\end{center}
\raggedright \footnotesize  \vspace{0.1cm} 
\textit{\underline{Notes:}} The above table uses data from all rounds of the National Sample Survey between 1999-2005. Each outcome variable takes the value 1 if an individual reports working on an informal contract (Column 1), informal firm (Column 2), and casual work (Column 3). Daily log wages are reported in Column (4). Post is a binary variable that takes the value 1 for years after 2003, while Treat is a binary variable that takes the value 1 for Andhra Pradesh and 0 otherwise. All regressions contain state-2 digit industry and 2-digit industry-year fixed effects along with a state-time trend. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. The sample size varies depending on the coverage of the underlying variable across rounds as well as whether a worker reports it. We do not use sampling weights.
\end{table}









%Table A9
\begin{table}[ht!]
	\caption{Excluding Self-Employed Individuals from the NSS Analysis \label{tab:nss_woself}} 
\begin{center}
	\input{nss/TableA9}
\end{center}
\raggedright \footnotesize \vspace{0.1cm} 
\textit{\underline{Notes:}} The above table uses data from all rounds of the National Sample Survey (NSS) between 1999-2005. Each outcome variable takes the value 1 if an individual reports working on an informal contract (Column 1), informal firm (Column 2), and casual work (Column 3). \autoref{tab:datadef} provides detailed definitions for each outcome variable. Panel A includes self-employed individuals, who are classified as informal firms, and casual workers. Panel B excludes them. Post is a binary variable that takes the value 1 for years after 2003, while Treat is a binary variable that takes the value 1 for Andhra Pradesh and 0 otherwise. All regressions contain state-2 digit industry and 2-digit industry-year fixed effects along with a state-time trend. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. The sample size varies depending on the coverage of the underlying variable across rounds as well as whether a worker reports it. The regressions include the NSS sample weights. 
\end{table}









%Table A10
\begin{table}[ht!]
	\caption{Alternate Definitions of Informal Workers \label{tab:nss_rob_inf}} 
\begin{center}
	\input{nss/TableA10}
\end{center}
 \raggedright \footnotesize  \vspace{0.1cm}
\textit{\underline{Notes:}} The above table uses data from multiple rounds of the National Sample Survey. Columns (1) and (2) restrict the sample to the main sample years between 1999-2005, whereas Columns (3) and (4) include all available years. The outcome variable takes the value 1 if an individual reports working on an informal contract, which is defined to be wages received on a daily or weekly basis in Columns (1) and (3), and daily basis in Columns (2) and (4). Post is a binary variable that takes the value 1 for years after 2003, while Treat is a binary variable that takes the value 1 for Andhra Pradesh and 0 otherwise. All regressions contain state-2 digit industry and 2-digit industry-year fixed effects, and a state-time trend. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. 
\end{table}








%Table A11
\begin{table}[ht!]
	\caption{Event Study using the NSS \label{tab:nss_eventstudy}} 
\begin{center}
	\input{nss/TableA11}
\end{center}
\raggedright \footnotesize
\textit{\underline{Notes:}} The above table uses data from multiple rounds of the National Sample Survey between 1993-2009. The outcome variable in daily log wages are reported in Column (1), and takes the value 1 if an individual reports working in casual work in Column (2). Treat is a binary variable that takes the value 1 for Andhra Pradesh and 0 otherwise. All regressions contain state-2 digit industry and 2-digit industry-year fixed effects, along with a state-time trend. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. The sample size varies depending on the coverage of the underlying variable across rounds as well as whether a worker reports it.
\end{table}







%Table A12
\begin{table}[ht!]
	\caption{Heterogeneity by Manufacturing and All Sectors \label{tab:nss_manf}} 
\begin{center}
	\input{nss/TableA12}
\end{center}
\raggedright \footnotesize  \vspace{0.1cm} 
\textit{\underline{Notes:}} The above table uses data from multiple rounds of the National Sample Survey between 1999-2005 for the main sample, and between 1993-2009 for the ``all years'' sample. Each outcome variable takes the value 1 if an individual reports working on an informal contract (Column 1), informal firm (Column 2), and casual work (Column 3). Daily log wages are reported in Column (4). Panels A and B restrict the sample to the manufacturing and all sectors for the main sample years, while Panels C and D use all years available. Post is a binary variable that takes the value 1 for years after 2003, while Treat is a binary variable that takes the value 1 for Andhra Pradesh and 0 otherwise. All regressions contain state-2 digit industry and 2-digit industry-year fixed effects. Panels C and D additionally include a state-time trend. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. The sample size varies depending on the coverage of the underlying variable across rounds as well as whether a worker reports it.
\end{table}






%Table A13
\begin{table}[ht!]
	\caption{Impact on LFP, Unemployment, and Agricultural Work \label{tab:nss_unemp_ag}} 
\begin{center}
	\input{nss/TableA13}
\end{center}
\raggedright \footnotesize \vspace{0.1cm} 
\textit{\underline{Notes:}} The above table uses data from multiple rounds of the National Sample Survey between 1999-2005. Each outcome variable takes the value 1 if an individual reports working in the labor force (Column 1), is unemployed (Column 2), or works in agriculture (Column 3). Post is a binary variable that takes the value 1 for years after 2003, while Treat is a binary variable that takes the value 1 for Andhra Pradesh and 0 otherwise. All regressions contain state-2 digit industry and 2-digit industry-year fixed effects, and a state-time trend. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. 
\end{table}



%Table A14
\begin{table}[ht!]
\caption{Robustness to Alternate Definitions of Large Firms \label{tab:ec_firmlevel_het}} 
\begin{center}
\centering
\input{ec/TableA14}
\end{center}
\raggedright \footnotesize \vspace{0.1cm}
\textit{\underline{Notes:}} The above table uses data from the 1998 and 2005 rounds of the Economic Census. The outcome variable is a binary variable that takes the value 1 if a firm is unregistered and 0 otherwise. We classify firms as Large if they employ more than 10 workers (Column 1), 20 workers (Column 2), 30 workers (Column 3), 50 workers (Column 4), and 100 workers (Column 5). All regressions include district, 2-digit industry, state-industry, and year fixed effects, along with controls for firm characteristics like whether the firm is owned by a female, whether the owner is from a lower-caste, and whether the firm uses electrical power. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. 
\end{table}

%Table A15
\begin{table}[ht!]
\caption{Robustness to Restricting Results to the Manufacturing Sector Only \label{tab:ec_firmlevel_manf}} 
\begin{center}
\centering
\input{ec/TableA15}
\end{center}
\raggedright \footnotesize \vspace{0.1cm}
\textit{\underline{Notes:}} The above table uses data from the 1998 and 2005 rounds of the Economic Census. We restrict the firms to the manufacturing sector only. The outcome variable is a binary variable that takes the value 1 if a firm is unregistered and 0 otherwise. We classify firms as Large (Column 1) if they employ more than 10 workers, and Small (Column 2) if they employ less than 10 workers. Column 3 includes all firms. All regressions contain district, year, and state-2-digit industry fixed effects and control for firm characteristics like whether the firm is owned by a female, whether the owner is from a lower-caste, and whether the firm uses electrical power. Control, Pre Mean reports the mean of the outcome variable among the control states in the years before the policy reform. Robust standard errors are clustered at the state level and reported in parentheses below. * is $p<0.1$, ** is $p<0.05$, *** is $p<0.01$. 
\end{table}


%Table A16
\clearpage \newpage
\begin{table}[ht!]
	\caption{Correlation between Firm-Size and Fraction of Contract Workers}\label{tab:asi_estimation}
	\begin{center}
	\centering
	\input{calibration/TableA16}
	\end{center}
	\raggedright \footnotesize \vspace{0.1cm}   
	\textit{\underline{Notes:}} The above data is at from the Annual Survey of Industries for Andhra Pradesh in 2002. Column (1) reports the regression estimates without any fixed effects, while Column (2) adds 2-digit industry fixed effects. Robust standard errors are reported in parentheses. * is $p < 0.1$, ** is $p < 0.05$, *** is $p < 0.01$.
\end{table}



%Table A17
\begin{table}[ht!]
	\centering
	\caption{Model Fit}\label{table:Model_Fit}
	\begin{center}
		\input{calibration/TableA17}
	\end{center}
	\raggedright \footnotesize \vspace{0.1cm}
	Notes: The above table reports the moments in the model (Column 1) with those in the data (Column 2).
\end{table}



%Table A18
\begin{table}[ht!]
	\centering 
	\caption{Derivatives of Moments to Parameter Changes}\label{table:Derivatives}
	\begin{center}
	\input{calibration/TableA18}
	\end{center}
	\raggedright \footnotesize
	\textit{\underline{Notes:}} This table reports the derivatives of each moment with respect to each parameter. Each row is a moment calculated from the model simulation. Each number in the table indexed by row R and column C, is the percent change in the moment in row R, when a parameter in column C is increased by 1 percentage.
\end{table}



	
\end{document}